Sarvam becomes India's newest AI unicorn on a $234M HCLTech-led round
A $1.5B valuation built on a sovereign-AI pitch, sharpened by Anthropic cutting off foreign access to its top models
أضف إلى قائمة
لا قوائم بعد.
Summary
Sarvam AI closed the first $234M of a $300M Series B on 15 June 2026 at a $1.5B post-money valuation, making the Bengaluru startup India's newest AI unicorn. Lead Hcltech, the IT arm of the HCL Group, put in $150M; Bessemer Venture Partners joined alongside existing backers Khosla Ventures and Peak XV Partners. Sarvam builds models tuned for Indian languages and is deploying across banking, insurance, government services and defence. It will fund work on next-generation agentic, coding and cybersecurity models and expand compute access. The timing is pointed: it lands days after Anthropic disabled foreign-national access to its Fable 5 and Mythos 5 models on a US government order, hardening the case for home-grown frontier models.
The split
US venture coverage (TechCrunch) reads the round through the global sovereign-AI lens and the Anthropic cutoff. Indian outlets (Business Standard) treat it as a domestic policy test: can a private IT conglomerate, not the state, anchor a national model stack, and does "sovereign AI" survive contact with commercial deployment in banks and ministries. The framing diverges on the actor: Silicon Valley sees a hedge against US export control; Delhi sees an industrial-policy proof of concept.
By the numbers
- $234M, first close of a $300M Series B.
- $1.5B, post-money valuation.
- $150M, HCLTech's lead cheque.
- 130th, India's unicorn count after the round.
- 4+ sectors named: banking, insurance, government, defence.
Why it matters
Frontier-model access is concentrating in a few US labs that can be ordered to cut off foreign users, which turns "sovereign AI" from a slogan into procurement strategy for India and others. Sarvam is the clearest Indian bet that a domestically controlled stack, backed by a strategic conglomerate, can serve regulated sectors that cannot depend on US models.
What to watch
- The second close completing the $300M target, and who else joins.
- Whether Sarvam's models win Indian government and defence contracts on the sovereignty pitch.
- Compute sourcing, can it secure GPUs at scale under tightening export rules.